Biological agents pursue their goals in a lifelong series of encounters with noisy and changing natural environments, relying on incomplete and costly information about these environments. In such fundamentally uncertain natural contexts, learning and decision-making are inherently coupled in two ways. First, learning, understood as the updating of representations that serve as inputs to decisions, must ultimately be in the service of better decision-making: successful learning enables better decisions by appropriately reducing decision uncertainty. Second, agents must continually make decisions about the appropriate allocation of scarce cognitive resources, including how, at any given time, to weigh the value of learning itself against the values of competing goals, such as maximizing near-term extrinsic rewards (explore-exploit tradeoff). Consequently, learning is both necessary for effective decision-making under uncertainty (constitutive aspect), and is itself an object of decision-making (allocative aspect). A goal of our research is to understand how the human brain uses representations of uncertainty to inform these two aspects of the interplay between learning and decision-making.